#codefresh (2020-03)

I’m trying to build 2 images. The second one depends on the first one. I’m trying to pass NGINX_IMAGE=${{build_nginx}} as a build argument, but the value comes in blank. Is there something that I’m missing here, or do I have to construct the image name myself (which I would prefer to avoid).

That’s a special sha that is understood by the pipeline and not something that would work outside our our image: construct. Let me look into a couple things and come back to you on this Harrison.

Thanks.

I would think it should work for anything though, because really it should just represent the image name.

Here is the variable you are likely looking for but even that will possibly need some modifications ${{build_nginx.imageId}}

Gotcha, I will give it a shot.

That at least prints image information. I am looking over some other details.

Example output: [r.cfcr.io/dustinvanbuskirk/dustinvanbuskirk/rails-app-with-knapsack_pro-test@sha256//r.cfcr.io/dustinvanbuskirk/dustinvanbuskirk/rails-app-with-knapsack_pro-test@sha256:f65847a0ee9f868582a7b36a5602077c6708414a72ac71a679472f1a10939b9c)

That link doesn’t work but I am giving this a shot myself

Sorry Slack just made it a link…

It’s the output of the imageId from one of my builds.

Oh I gotcha.

I would think that should work. Thanks for the quick answer. The pipeline is running right now.

Here is the tag example.

Executing command: echo ${{BuildTestDockerImage.tag}}             
master-a6a1d3f                           

Oddly this did not work. I must have done something wrong.

Oh, the change didn’t push for some reason, that’s why

Not necessarily. Like I said that SHA is something internal to us.

I am checking the name as well so you can concat the two

It did push, I just forgot I had this in 2 spots, so I’m trying again

You might need to do NGINX_IMAGE=${{build_nginx.imageName}}:${{build_nginx.tag}}

Okay, hopefully this helps. Ping me using @dustinvb if this doesn’t get you where you need to be.

What’s weird is I am getting invalid reference format: repository name must be lowercase with both ways of doing it.

That is a little bizarre maybe with the second name we need repo information.

I’m running it again with a step to output what the value is for debug purposes.

[r.cfcr.io/linioit/shop-front-nginx@sha256:d19c2542673f9c191cdc66a43960052e4a65ad3b00718c69776778840d428d46](http://r.cfcr.io/linioit/shop-front-nginx@sha256:d19c2542673f9c191cdc66a43960052e4a65ad3b00718c69776778840d428d46) was the value

So this is weird

It works when I do it locally. Could it be the version of docker being used by venona?

That should be 18.09 last I checked. I assume the node Docker is the same or newer.

I would imagine so. We’re on GKE 15

Could CF be passing the value in caps for some reason?

You Dockerfile, if the first like FROM: ? or the Build Arg?

No it shouldn’t be doing that. I printed using echo those vars in my step. They were all lower case. The build arg should not change that case.

build_arguments:
      - NGINX_IMAGE=r.cfcr.io/linioit/shop-front-nginx@sha256:d19c2542673f9c191cdc66a43960052e4a65ad3b00718c69776778840d428d46

This worked

So it is 100% CF messing up the value when passing it in

Can you open a Zendesk that includes the Build ID of the build with the error about capitalization so we can both review the YAML and see the error?

Yeah.

Can you also share the build id with me over DM here so I can just quickly peek into the YAML?

Sure, one second

Thanks, I recreated and will update ticket here soon.

In the meantime can you try this as a workaround, let me know if you have better results by using a direct variable instead of the metadata in the build_arguments.

  ExportVar:
    image: alpine
    commands:
      - cf_export NGINX_IMAGE='${{BaseImage.imageId}}'
  BuildArg:
    title: 'Build image: php-test'
    stage: build
    type: build
    image_name: '${{CF_REPO_NAME}}-php-test'
    dockerfile: Dockerfile
    build_arguments:
      - GITHUB_TOKEN=${{GITHUB_TOKEN}}
      - NGINX_IMAGE=${{NGINX_IMAGE}}
    tag: '${{CF_SHORT_REVISION}}'

Is it possible in codefresh to authenticate before a docker build using the builtin build step?

I have a Dockerfile that uses a private image as a base image i.e.

FROM private/image/repo:v1.1.1

I’d like to be able to authenticate (docker login) for the build. I know I can do this in a freestyle step but I’d like to know if i can with the builtin build step.

Helm quick start guide for Helm 3 - documentation () (Feed generated with FetchRSS)

Helm 3 recommended labels - documentation () (Feed generated with FetchRSS)

Helm 3 release dashboard - documentation () (Feed generated with FetchRSS)

Deployment example for Helm 3 - documentation () (Feed generated with FetchRSS)

If you’re relying on the internal codefresh registry a lot (and I definitely am) you’ve got your work cut out for you:

https://codefresh.io/codefresh-news/deprecation-of-the-codefresh-private-registry/

1.5 month notice to change your repos

disappointing they would give such a short period of time

Also, no real way to see your usage of it and be able to validate you’ve actually migrated your stuff off of it

I am also using it for one of my private projects kinda sucks.

Better the devil you know….

Hi there: I wanted to update this channel here that we got an extension from our partner to keep CFCR active for longer

We are finalizing the dates now and will announce. But we have at least until July for now

Thank you for your patience with us while we go through this.

Thanks @Vidhya Vijayakumar for the update

@Vidhya Vijayakumar will codefresh provide a way to test or validate that a pipeline is not affected?

We are working on a migration guide to help everyone. I have requested this already with our team. But, it may not be in the first document released

We will definitely push updates

@Vidhya Vijayakumar Did something change with the ${{build.imageId}} today? Because our builds have started failing. It no longer outputs the repository.

Previous: [r.cfcr.io/linioit/shop-front-nginx@sha256//r.cfcr.io/linioit/shop-front-nginx@sha256:a424b6cf529d897f35ea172ca502ba9ebb4e4ab6bdf6aac9c0314508c14782ff) Current: sha256:b9fa6fd9b0724f8b26da274e36c58886aabd37eb69fd7c3ba0b7678a169121cb

Environment board now supports Helm 3 - documentation () (Feed generated with FetchRSS)

Before I open a ticket, was hoping someone might have some insight on this error I’m getting. This is during the first step of my pipeline when pulling the Docker image to run the step with:

failed to register layer: Error processing tar file(exit status 1): write /var/lib/helm/cache/plugins/https-github.com-mstrzele-helm-edit/vendor/git
hub.com/gobwas/glob/glob_test.go: no space left on device      

I’m assuming there is some cleanup that isn’t happening and I’ve filled the mounted volume that exists within each codefresh pipeline, but not sure best route to handle that. Unfortunately, a few of my images are rather large, around 1GB, and so are some of my git repos. I just started running in to this though on pipelines that have been working, so I’m wondering if there is some docker build cache that’s not being cleaned up or something.

It looks like the 1.x version of venona was released, but I don’t see any documentation on upgrading. Nor do I see any way to install this with regular manifests or a helm chart (preferable)

Removal of the Codefresh registry - documentation () (Feed generated with FetchRSS)

Adding @discourse_forum bot

@discourse_forum has joined the channel